Egocentric Video Data Collection: Daily Household & Commercial environments Activities (Remote)

We are seeking individuals or vendor partner to participate in an egocentric video data collection project.Awign runs large-scale, factory-grade egocentric (first-person) video and sensor data collection used to train embodied-AI and robotics models. We are expanding delivery into Latin America .

(Total expected volume is up to 1 million QC-approved hours. ) - Duration of Project 6 month

What You Will Do:
Participants will be required to wear an iPhone on their head using a provided mount to record their daily activities.
1.Egocentric footage:
•  first-person video of real tasks performed by workers or you, captured with a head- or
forehead-mounted device(iphone) angled downward so both palms and all fingers stay visible while manipulating
objects.
• Privacy guardrails: recording only during active task performance; no personal faces or ID information in
frame; footage never shared publicly; documented consent obtained at every site. These are non-negotiable.
• What counts: we are paid on QC-approved hours, not raw recorded hours. Quality is built into the process,
not inspected at the end.
2. Two environments
Residential is split evenly across four task families; commercial spans a wider set of real workplaces.

A) Residential — equal split across task families (25% each)
1.Laundry tasks
2.Kitchen tidy-up (not cooking)
3.Organising a single room
4.Other household tasks

B) Commercial — real workplace environments
1.Cloud kitchens
2.Hospitality & housekeeping
3.Packaging
4.Waste management
5.Home construction
6.Skilled trade
7.Field-services light installation
8.Warehouses
9. Confectioneries
10.FMCG repacking & white-labelling
11.Supermarket / retail
12.Handmade production
13.Workshop
14.Other approved categories

3. Capture devices & technical bar
• Accepted phones: recent high-end smartphones — iPhone 12 and newer, Google Pixel 6 and newer,
Samsung Galaxy S21 and newer. No other phones unless a compliant external camera is wired to the phone.
•Aspect Ratio:1:1 (square) OR 16:9 (landscape – preferred) -Videos should be recorded in landscape mode
•Clip structure: clips bounded by complete semantic tasks (e.g. One full 29-min task as a single clip),
minimum 120 seconds per task.
4. What fails QC
• Obstructed, tilted, shaky or poorly lit footage; contrived or robotic movement; hands held up to the camera to
start a recording.
• Hands out of frame for more than 10% of an active grab/release; participants seated; unnecessary pauses;
extended repetitive motion (e.g. Polishing); any footage without proper consent.
5.Commercials
• Data collection payout: usd 4 per qc-approved recording hour. (Payment will be released only after successful Quality Check (QC) validation)
• Onboarding: begin small to prove quality and consent processes, then scale to full volume; operations within
~24 hours of equipment readiness. Awign provides consent templates, training manuals and ongoing support.